Bio Inspired Computing for Outlier Detection Select Studies in Web 3 0 Domain

Abstract

Data analytics has emerged as an inevitable domain. Increasing magnitude of data not only in terms of volume but also variety and veracity has made the subsequent analysis and decision making a challenging task. Researches and practitioners have adopted variety of data analytics approaches and frameworks for retrieving useful information from data of such magnitude. The entire business intelligence can actually go futile if the available data is not in the correct format or comprises of aberrations/outliers. These data instances may occur due to errors made while acquiring the data, data variations or some deviations in the data itself that result into abnormalities. This makes outlier detection an inevitable step for efficient and effective information retrieval. Further, advances in the domain of information technology have increased exponentially with the rising growth in the use of the internet gradually generating innovation in diverse domains. This leads to the emergence of Web 3.0 with huge amount of data being generated from social media and other interactive web platforms. Thus, the contribution of this work is twofold, both methodological as well as application oriented focusing on the domain of Web 3.0. Methodologically, the work proposes several hybrid bio-inspired computing algorithms by integrating them with traditional algorithms. The bio-inspired computing algorithms are known to produce promising results when compared to traditional machine learning algorithms that are usually utilized for outlier detection. The select studies use the proposed hybrid approaches for outlier detection in relevant studies of Web 3.0. The work is focused on three research problems in the Web 3.0 domain including search engine marketing, social media marketing and influencer marketing. The use of hybrid bio-inspired computing algorithms eliminates locally optimum solutions and catalyzes the convergence of the solution. newline newline newline